See exactly what changed
Compare two scored runs of the same agent, or the pass rates your scans recorded for each model, module by module. Only runs graded by the same corpus and grader versions are compared.
In this release
You can use this in your workspace today.
- Scan compareCompare scored runs of the same agent and follow changed findings back to their evidence.Live
- Model comparisonCompare recorded model pass rates by module, with graded coverage and evidence.Live
Run against run
Pick two scored scans of the same agent and see which findings are new, which are fixed and which are unchanged, each linked to its evidence.
Score and grade change
See how the security score and grade moved between the two runs, from the same graded tests.
Model against model
Compare the pass rates your scans recorded for each model, by module, with how many tests were graded behind each rate.
Comparable or refused
Runs graded by different corpus or grader versions are not compared, and a module with nothing graded shows no rate rather than a zero.
Run two scans
Assess the same agent twice with the same profile, before and after a change.
Compare
Open the comparison from the scan history and choose the two runs.
Follow the evidence
Open any changed finding to read the request, the response and the rule behind it.